AI Readiness Sprint
2 weeks
You'll know which two or three AI use cases are worth your money, which aren't, and exactly what to do in the next 90 days - with a case your board can sign off.
The gap
Mid-market firms want AI. Almost nobody sells to them properly.
If any of this sounds like your last twelve months, the problem isn't the technology.
01
Overwhelmed by hype
Every vendor promises transformation. Few deliver an outcome anyone can measure. Knowing where to start has never been harder.
02
Locked out by price
The consultancies with the strongest practitioners charge enterprise day rates - and you pay for associate teams and slide decks, not senior delivery.
03
Burned once already
The software went live and nobody used it. Processes unchanged, ROI unproven. The tool was fine. Adoption was the problem.
How we can help
Named work with a defined deliverable - not open-ended day rates.
Each one ends with something you keep and can act on - not a report that sits in a drawer. Fixed timescale, written price before you commit to anything.
Copilot Launchpad
4 weeks
Your team using Copilot properly rather than ignoring it - with a policy that keeps you safe, prompts built around your actual work, and numbers showing who has adopted it.
Process Automation Pilot
6 weeks
One process that currently eats your team's week running itself - and your own people able to keep it running after we've gone.
Data Foundation Review
3 weeks
A straight answer on whether your data can support what you want to build, what's wrong with it, and what to fix first - before you spend anything on tooling.
AI Governance Pack
3 weeks
The ability to answer a client, insurer or regulator asking how you control AI - and a framework that holds up when a question arises you didn't anticipate.
Fractional Digital Director
Ongoing
A senior technology voice in the room when the decisions get made - so you stop buying tools you don't need and finish the ones you've started.
Not sure which one? The Readiness Sprint exists to answer exactly that, in two weeks.
How engagements grow
Start small. Only scale what works.
Most clients begin with a two-week assessment and decide from there. Nothing commits you to the next stage.
Assess
Find out what's actually worth doing, before spending anything on doing it.
- Digital and AI maturity baseline
- Process audit
- Prioritised roadmap
Transform
Senior-led delivery, built so your own team can own it afterwards.
- Power Platform builds and Copilot deployment
- Process redesign
- Change, training and adoption
Sustain
A trusted digital adviser on call, so the value doesn't decay after go-live.
- Quarterly strategy reviews
- New use cases as they emerge
- Platform health and governance
Where this goes deepest
If your operation runs on documents, that's where the money is.
In regulated and professional firms, the expensive work isn't decision-making - it's the reading, checking, re-keying and chasing that surrounds it. That work is invisible on the org chart and enormous on the payroll.
Reporting packs assembled by hand from a dozen sources. Onboarding files checked twice because nobody trusts the first pass. Policy documents, valuations, statements and contracts moving between people who each extract the same fields again.
This is the highest-return application of AI in a mid-sized business, and also the one most often done badly - because a generic assistant bolted onto a document-heavy process produces plausible output nobody can rely on. Getting it right needs the process mapped before anything is built, and validation designed in from the start.
Typically
- Accountancy and audit firms
- Legal practices
- Funds and investment managers
- Insurers and brokers
- Lenders and finance providers
01
Map the process first
Before any build: the real workflow, including the handoffs and the exceptions people work around quietly. Most document processes aren't what the procedure manual says they are.
02
Own the whole path
From intake to the point a decision is made - not a tool dropped in the middle for your team to integrate. A partial automation usually adds a step rather than removing one.
03
Design for verification
Structured extraction with a human checkpoint where the consequences justify one. In regulated work, output nobody can audit is worse than no automation at all.
3 weeks
Fixed scope
Document Workflow Map
A costed picture of what one document process really consumes in hours and handoffs, and which parts are worth automating. Useful on its own, and the basis for a build if the numbers justify one.
The Mantra Method
Prove it here before you scale it here.
The commonest mid-market failure is a large deployment decision taken on a vendor's claims rather than evidence from your own environment. This method makes that impossible.
1
Orient
Where are we actually?
Maturity baseline
2
Prioritise
What's worth doing first?
Ranked use-case register
3
Prove
Does this work here?
Pilot in production
4
Scale
How do we roll it out?
Governed deployment
5
Sustain
How does value survive us?
Runbook and ownership
Sustain returns to Orient. The method is a cycle, not a line - a fresh baseline twelve months on, because transformation isn't an event you finish.
AI governance
"Can we?" is an engineering question. "Should we?" is not.
Most AI governance is written to satisfy an auditor. It produces a policy nobody reads and a risk register nobody revisits. Useful governance starts earlier than that.
We bring a second discipline to this work: spiritual intelligence - the capacity to hold questions of purpose, proportion and consequence alongside questions of capability. It isn't a religious idea, and it isn't soft. It's the oldest part of the management canon, and the part the technology industry has been most willing to skip.
Purpose
Before capability
A process that can be automated doesn't have to be. Which judgements should stay human, and for what reason you'd be willing to defend in public.
Proportion
Right measure
Where automation genuinely serves people, and where it quietly hollows out the work that made them good at their jobs. Both failures are expensive.
Stewardship
Who bears the cost
Decisions taken now bind staff, clients and successors who had no say in them - and carry a real compute, energy and organisational cost.
Accountability
A named human
Someone must be able to explain a decision and own it. Responsibility that dissolves into a model is not governance, whatever the policy says.
Where this thinking comes from
The vocabulary is old. Dharma - duty proportionate to your role and circumstance, rather than to your capability. The Aristotelian mean - virtue found between deficiency and excess, which is exactly the judgement an automation decision requires. Stewardship - the obligation to hand on what you were trusted with in better condition than you found it.
These aren't decorations on a compliance exercise. They're the questions a board actually wrestles with once the technology stops being theoretical and starts affecting people they employ. A governance framework that can't engage with them produces documents, not decisions.
The practical output is the same either way: a usage policy, a risk register, regulatory mapping and guidance your staff can follow. What changes is whether the framework underneath it can answer a hard question you haven't anticipated yet.
3 weeks
Fixed scope
AI Governance Pack
Usage policy, risk register, EU AI Act and UK regulatory mapping, and plain-English staff guidance - built on a board session that settles what your organisation actually wants to stand behind, and a "should we" test applied to every use case on your register.
Nilesh Solanki
Founder & Principal
- 25 yrsDigital transformation inside a Big Four firm
- 1,400+UK practitioners in the Power Platform community he led
- 100+Applications delivered through the Centre of Excellence he founded
- £50m+Revenue delivered on a Secure Data Service programme
- BoardsTrustee roles across national charities and a schools trust
Who you work with
The person you meet is the person who does the work.
Nilesh spent twenty-five years inside a Big Four firm, where he founded and scaled its national Power Platform Centre of Excellence from a standing start - governance, adoption, training and delivery for a community of over a thousand practitioners.
That experience taught us something the vendor market keeps getting wrong: the technology is rarely the hard part. Getting several hundred people to change how they work, safely and permanently, is the hard part. It's also the part that determines whether the investment returns anything.
Mantra exists to bring that discipline to businesses that could never justify a Big Four engagement - at a fifth of the cost, with no leverage pyramid and no junior team learning on your budget.
Where a build needs specialist depth - heavy document extraction, model work, sector-specific engineering - we bring in a delivery partner we've worked with and tell you exactly who they are before anything is agreed. The relationship, the judgement and the accountability stay with us.
Start here
Ask a question. It costs nothing and commits you to nothing.
Bring the problem you're actually stuck on. If we're not the right fit for it, we'll say so and point you somewhere better.
Most enquiries turn into a 30-minute call. If it fits from there: a 90-minute scoping session - free, and useful to you whether or not we work together - then a written proposal within five working days.
- Email nilesh@mantraconsultancy.co.uk
- Phone +44 7956 957864
- Based in London, working across the UK and internationally
Send an enquiry
Tell us what you're dealing with. Every enquiry is read by Nilesh personally, and answered within one working day.
This opens an email addressed to us with your details filled in - nothing is sent until you press send in your own mail app. No newsletter, no follow-up sequence, no passing your details on.